EXACTLY Theoretically could do it. Real world, no way. I do 3pt goal/ 1.5 risk. In reality, this usually equals about 3:1 cause I slowly move the stop. By the end of the month it averages something like 2.25/ .75. works well. When I try for anything more than 4 pt goal, I'm wrong on a very high percentage. I'm not a guru, but I'm pretty good. I can't fathom someone making 50% on a 6 point goal daytrading ES. Jay
this is what i meant by saying earlier that each change in position can only be a reflection of changes in the market and not just a function of my pnl. the account might serve as an indicator for what happened in the market place. but it cannot have any value by itself.
btw i think acrary is the only person who can create lasting threads without really participtating ...
There is no need to restate something that has been already addressed. But I do think that this following point from Cutten did not get the proper attention it deserved, so I'm bringing this back from the dead as well. Many would do well to understand Cutten's observation. Position sizing and expectancy alone do NOT constitute as an edge. To put it simply, having tighter stops and tighter profit targets will result in higher win rates, but at the expense of the 'fat tails' or large winners acrary mentioned. Conversely, having looser stops and larger targets will result in a lower win rate, but higher average wins. Remember that expectancy is calculated: (Average win * %tage winners) - (Average loss * %tage losers) Manipulating target R/R profiles will just allow one to tailor trades to their own personality and preferences for win/loss rates. Cutten, your following assertions here contradict your point above. Statitically, the 3:1 system is inherently edgeless. As a result, it doesn't matter if you're trading 2min charts or weekly charts. The R/R profiles will just be respectively scaled to the proper fractal. Theoretically, one would be able to use this to set very tight stops to lower the potential risk but also the potential return. This might result in higher loss rates though. Acrary does assume a 50% win rate for this proposal, but in actuality this can vary depending on the trading system used. Generally in practice one will find about a 25% win rate about the right rate for a 3:1 R/R for a 0 expectancy: (3 * .25) - (1 * .75) = 0. Note how you can still "break even" with such a low win rate. If you have a better win rate than 25%, you will actually have a positive expectancy and make money. Any other result than the 25% win rate mentioned above in practice will be due to a system's edge or market conditions, or a combination of both factors. Note that these two factors are NOT the same, although they each play a role in the outcome of win and loss rates. Another insightful observation by Cutten. You are correct that two traders having the same positions in a vehicle will experience the same outcome going forward, regardless of their respective entries. This is because, obviously, the underlying is the same. However, the risk/reward for each trader is in fact different. Using the definitions of risk put forth in this thread so far, say Trader A got into a stock at $10 with a stop of $9 (risk = 1) and a target exit of $13 (reward = 3) and Trader B got into the same stock at $12 with a stop of $9 (risk = 3) and a target exit of $13 (reward = 1). See how one trader has 1/3 risk/reward while the other has 3/1? The only way that they have the same R/R would be if they placed the trade at the same entry point and have the same stops and target exits. I think what you are getting at here is that a new measure of risk must be used to define a trade, after entry. I stumbled upon this dilemma some time ago as well, and came up with a rather simple solution: distinguishing initial portfolio capital risk from open portfolio capital risk. I call the former Initial Capital Risk or ICR and the latter Net Asset Value Risk or NAVR. ICR is calculated as Target Entry Price - Stop Price. NAVR is calculated as Current Price - Stop Price. Once a trade starts moving, stops must be adjusted dynamically, altering the initial R/R profiles. Typically, the risk will shrink as the trade progresses in your favor and you move the stop, and the reward will remain, allowing one to pyramid on positions while keeping the same risk, or just reducing risk in general while keeping the same potential reward. By calculating the aggregate NAVR for all positions in a portfolio, one can find the true NAVR for the whole portfolio - that is, if all of your positions got stopped out (disregarding slippage/commissions), how much of your NAV you would lose. In practice, one will find that the aggregate ICR, on the other hand, may actually be smaller than NAVR; typically this happens when a position you have has moved significantly and you've moved your stop at or past breakeven - you face the potential of a pullback, which will hurt your profit (NAV as of that time), but not your initial capital (IC). I find that keeping NAVR under a threshhold will limit portfolio fluctuations. Someone mentioned earlier that the whole point of this was becoming very selective in which trades one chooses to partake in. Without going into the nuances, I think for most, that would be the lesson to take away from all of this.
I just saw this reply. Your first point is mistaken - I did not say that acrary's position sizing gave any edge. Quite the opposite, i said it would give no edge in a random market. All I said was that *in a trending market* it would work well. The definition of a trending market is one that has a tendency to serial correlation, so if your first entry shows a profit then your trade expectation has increased and thus additional buys should show a profit. Your second remark also misses the point I made. If trader X enters at $10 with stop at $9, and trader Y enters at $12 with stop at 9...then when the asset is at $12, they both have identical risk/reward. When the asset has reached $12, it does not matter what the risk/reward was at $10. All that matters is the risk/reward at the current market price. There is no difference between initial capital and open position capital. Again this is demonstrated by my original example. If trader X has $10,000 and buys at $10, then sees the stock go to $12, he now has $12000. $2000 is "open profit" and $10,000 is "initial capital". Trader Y comes along with $12,000, and buys the stock at $12. Please explain now how Trader Y has a position that is in any way different from Trader X (I am assuming no taxes here, to simplify the argument)? They both own $12,000 worth of the same stock. If the stock rises to $13 they will both make $1000. If the stock falls to $10 they will both lose $2000. Their outcomes will be *identical* in all circumstances. So, how on earth can there be an distinction between "open profits" and "initial capital"? Losing $2000 is losing $2000, regardless of whether you made that $2000 from profit on a stock, or from working at your job. The money is still $2000 in both cases. Trader X and Trader Y, owning $12,000 of the same stock, have identical risk profiles and will make or lose identical amounts going forward. The only logical conclusion is that open profit vs initial capital is a false distinction. Capital is capital. By the same token, entry price is irrelevant for calculating risk/reward from the current moment in time. Risk/reward from this moment on is based on the market price now, not the market price when you entered. Therefore, any position sizing method which depends on entry price per se, is fundamentally flawed. Any position sizing method which would have Trader X act differently to Trader Y, despite having identical capital and identical positions, is fundamentally unsound. All risk control and sizing methods must judge the risk/reward and trade expectation from the current price only, regardless of where you entered, and regardless of how much "open profit" you are showing.
Cutten, thanks for responding. In theory, assuming all individuals are 100% rational and following maximum expectancy with indifference to drawdown, then yes you are correct. However, in practice, almost all people exhibit a utility curve when it comes to what can be referred to as 'winnings' and 'losses'. It has been alluded to as a weakness perhaps in constant-position sizing because, if you're in, say, a 20% drawdown, 5% loss means something very different than if you were, say, up 40%. Many position sizing strategies as such recommend trading 'smaller' to alleviate this; Even large funds find it difficult (if not impossible) to maintain or increase their holdings once a position has gone against them significantly, even though fundamentally/statistically the trade may have a higher probability of reverting.
Find a system that works at least 50/50. That's not that hard. The problem is sticking to it. It's better than nothing, that's for damn sure. I think this is a great idea. I personally use a form of trailing stop....however adding to a winning trade is MUCH more preferable. Always willing to try something different. I might give this a shot Monday.
This is a great thread. ^Just to add to the logic above- If you use statistical modeling on the time series you're looking at, and can identify auto correction in what appears to be the current trend, then the probability of you entering a trade during random walk could also be reduced. I think the understanding of why you add to winning positions is misunderstood by alot of traders. However it's definitely a common trait of the few who eventually find success. Once upon a time I used to enter trades that paid 1:1, and would never add to a winning position. I pity the trader I once was.